Unearthing the gems.
Posted May 1, 2007
Recorded calls between your agents and customers hold valuable insights into how business is being handled within your company. Although many hours of these calls are recorded, very few companies are leveraging the vast amount of BI hidden within these audio files. Not making the most of this intelligence can cause companies to miss key opportunities to impact profitability, not to mention overall productivity and customer experience. Speech analytics enables quick and easy searching of large volumes of recorded audio, generating insight into the content of each conversation and, through aggregation and analysis, translating these insights into a deeper understanding of the business's performance.
The impact of speech analytics can be realized in three major areas.
Companies are constantly striving to improve efficiencies by reducing cost and increasing productivity, without sacrificing quality of service. By accurately identifying and analyzing call types and the cost of servicing these calls, organizational processes and resources can be realigned to achieve the desired balance between the cost of doing business and the quality of customer service.
Business Intelligence and Market Research
The conversation between your customer and your agent contains nuggets of information that can be uncovered and analyzed for market and business intelligence. In these calls there is a wealth of raw information about your competitors, how your products and services are perceived, recommendations for new products and services, and much more. Speech analytics enables you to glean this information from your existing recorded assets and share this intelligence throughout the organization.
Quality of Customer Interactions
Customers continue to become increasingly discerning and demanding with regards to price, product, and quality of service. In addition, customers are keenly aware of threats to their personal details and demand privacy protection from the companies they choose to do business with. Much effort is put behind measuring and managing customer satisfaction in many areas; however, without speech analytics these measurements are often based on analyzing small samples of data manually, resulting in a largely abstracted and not very timely measurement of the customer experience. With speech analytics, all calls can be analyzed for myriad customer experience indicators and the results can be available within 24 hours or less.
What to Look For
In selecting a speech analytics application there are many points to consider.
The core technology behind the speech analytics application--there are two technology approaches, speech-to-text and phonetic. Speech-to-text relies on a dictionary to render the audio searchable. Phonetic search, on the other hand, uses phonemes (the smallest unit of human speech) to mark and search the audio files. These approaches carry several implications for your speech analytics implementation.
Scalability and cost of implementation--phonetic speech analytics is very computationally efficient, making audio searchable anywhere between 82 and 340 times real time. Speech-to-text is more resource demanding, taking between one and four times real time to generate the text representation of the audio.
Accuracy and relevancy--speech-to-text relies on a dictionary to accurately translate the recorded audio. The dictionary can be trained and modified to support your organizational vocabulary. However, any modifications to the dictionary mean that audio will need to be reprocessed to generate more accurate search results. Phonetic searches rely on the phonemes that form the words and phrases, meaning that everything is searchable without training or tailoring to your audio content.
Speed--usually measured as the speed of search after the audio is indexed. With speech-to-text, search speed is faster than phonetic search as it is simply a text search. However, phonetic search is incredibly fast; recent benchmarks show search speeds of over 500,000 times real time. What is important to note is pure search speed in itself is only part of the story--in reality the total time it takes for audio to be made searchable and searched is the true measure of speed and the true indication of the impact to your organization.
Language, accent, dialect, and quality of audio--when reviewing speech analytics solutions it is important to gain an understanding of how the solution can support all the accents and dialects, languages, and various qualities of audio your business generates. By far the best test is to see the solution index and search and analyze your audio. The phonetic approach is ideally suited to supporting accents, dialects, languages, and ranges of audio quality through its process of training on actual audio representative of these qualities.
Once a speech analytics solution is chosen, your organization will be in a much stronger position to better understand how you can strategically improve your business in many different areas--improved efficiencies, deeper business and market intelligence, and, ultimately, a more positive customer experience.
About the Author
Anna Convery is the senior vice president of marketing and product management at Nexidia. Please visit www.nexidia.com.
The company's latest offering looks to tie individual agent performance more closely with corporate initiatives.
Sponsored By: Jacada, Avaya, Confirmit, inMoment and BoldChat
Sponsored By: Genesys, Avaya, Verint, and Aspect